Single-well tracer push-pull method development for subsurface process characterization
Early-time tracer injection-flowback test for stimulated fracture characterization, numerical simulation uses and efficiency for flow and solute transport
by Shyamal Karmakar
Date of Examination:2016-06-15
Date of issue:2016-12-07
Advisor:Prof. Dr. Martin Sauter
Referee:Prof. Dr. Martin Sauter
Referee:Dr. Iulia Ghergut
Referee:Prof. Dr. Gunter Buntebarth
Files in this item
Name:Thesis_Shyamal.pdf
Size:6.54Mb
Format:PDF
Description:PhD Thesis
Abstract
English
Geological inherent knowledge, hydraulic test and geophysical methods can estimate most of the stimulated georeservoir properties. The transport effective parameters such as fracture aperture and effective porosity cannot estimate by using these methods. The in-situ methods, the inter-well test or single-well test, are sensitive to transport effective parameters. Transport effective parameters determine geo reservoir's efficiency, sustainability or lifetime. The inter-well test needs more than one well which is not typical to install during early stage of geo-reservoir development to avoid too much investment before a proven use is confirmed. Hence, single-well test design for a specific sensitivity regime from ‘early’ to ‘very-late’ pull signal is rather practical for transport effective parameter estimation. Moreover, a typical single-well test that design for ‘mid’ to ‘late-time’ signal also loaded with many sensitive parameters. Secondly, tracer test design, and parameter sensitivity estimation depends on numerical simulation reliability. Finite element and finite difference code based different numerical method shows significant improvement toward this parameter inversion and test design. The use of single-well (SW) short-term tracer signals to characterize stimulated fractures at the Groß-Schönebeck EGS pilot site is studied in chapter 2, part 1. Short-time tracer flowback signals suffer from ambiguity in fracture parameter inversion from measured single-tracer signals. This ambiguity arises commonly due to a certain degree of interdependence between parameters such as fracture porosity, fracture thickness, fracture dispersivity. This ambiguity can, to some extent, be overcome by (a) combining different sources of information, and/or (b) using different types of tracers, such as conservative tracer pairs with different diffusivities, or tracer pairs with contrasting sorptivity on target surfaces. Fracture height is likely to be controlled by lithostratigraphy while fracture length can be determined from hydraulic monitoring (pressure signals). Since the flowback rate is known during an individual-fracture test, the unknown parameters to be inferred from tracer tests are (i) transport-effective aperture in a water fracture or (ii) fracture thickness and porosity for a gel-proppant fracture. Tracers with different sorptivity on proppant coatings and matrix rock surfaces for gel-proppant fractures and tracers with contrasting-diffusivity or -sorptivity for a water fracture were considered. This simulation study has produced two significant results: (1) water fracture aperture can be effectively evaluated based on early-time tracer signals of a conservative tracer; and (2) by using the combination of matrix sorptive and proppant sorptive tracers, it is possible to estimate fracture thickness and porosity in gel proppant fractures from a single test. The injection and flowback of a small fluid volume, and thus little dilution of the injected tracers, has three practical advantages: (1) there is no need to inject large tracer quantities; (2) one does not have to wait for the tails of the test signals; and (3) the field and laboratory monitoring of the tracer signals does not have to be conducted for ultra-low tracer concentrations, which is known to be a major challenge, with the highly-mineralized and especially high organic content fluids typically encountered in many sedimentary basement georeservoir. Additionally, it requires only a very small chaser injection volume (about half of fracture volume). Short-term flowback signals from injection-flowback tracer test face a certain degree of ambiguity in fracture parameter inversion from the measured signal of a single tracer. To improve the early-time characterization of induced fractures, of either gel-proppant or waterfrac, we recommend using tracers of contrasting sorptivity to rock surfaces, and to proppant coatings where applicable. The application that described in early time flowback tracer test study article at chapter 2, part 1. However, the tracer was not exhaustively demonstrating its complete range of uses for stimulated georeservoir. Sorptive tracer either on proppant or on a matrix that used for stimulated fracture characterization has raised the question about the range of sorptive tracer to produce for an effective tracer test. For the purposes, a lower sorptive tracer than its minimum necessary was suggested and a sensitivity improvement factor (ratio between sorptive tracer signal changes to conservative tracer signals changes, s/c) approximately equal to √ (1 + 0.7× sorption coefficient, κ) is formulated. One needs to note that the higher the tracer's retardation, the lower is its fracture invasion, and consequently a poorer capability for characterizing the fracture as a whole. In principle, this could be compensated by increasing the chaser volume (i.e., by injecting sorptive tracers earlier than conservative tracers). Modeling flow and solute transport become a state of the art for a set of engineering and hydrogeological applications. For hydrogeological modeling, a number of numerical software is available as commercial code as well as many research initiatives is emerging to develop a new one. This section (Chapter 3) of the thesis attempts to develop solute transport module in fracture using COMSOL. FEFLOW software with it discrete feature element (e.g. fracture) module it can simulate fully couple process for flow, solute, and heat simulation. For the study of early time, tracer flowback signal, the flow, and solute transport process coupling in fracture-matrix domain is studied using tetrahedral mesh. To compare the consistency of numerical result with spatial and temporal discretization as well as in different numerical approach, a same numerical model set up in COMSOL. Qualitative comparison of the between the codes reveals that dispersivity tensor application can cause a minor variation in the tracer breakthrough in single-well tracer flowback simulation. The result is compared in terms similarities and capturing the spikes of injection and flowback in early flow back tracer test. A set of well-established software, frequently used for modeling flow and transport in geological reservoirs, is tested and compared (MODFLOW/MT3DMS, FEFLOW, COMSOL Multiphysics and DuMux). Those modeling tools are based on different numerical discretization schemes i.e. finite differences, finite volumes and finite element methods. The influence of dispersivity, which is directly related to the numerical modeling, is investigated in parametric studies and results are compared with analytical approximations. At the same time, relative errors are studied in a complex field scale example. For 1D and 2D cases all three tested modeling software show good agreement with the analytical solutions. By refining the grid discretization all four software packages get an improvement in accuracy. It is shown for the 2D problem that COMSOL Multiphysics needs a finer mesh to produce the same accuracy as FEFLOW and DuMux. For transport simulations in forced gradient, where a commonly expected dispersion or higher value occurs, the finite element software FEFLOW is the best choice. From this comparative study, it is revealed that under forced gradient conditions, finite element codes COMSOL and FEFLOW show a higher accuracy with respect to the analytical approximation for a certain range of dispersivity than DuMux and MODFLOW/MT3DMS. Comparing simulation time and code parallelization, FEFLOW performs better than COMSOL. Computational time is lowest for finite difference software MODFLOW/MT3DMS for a small number of mesh elements (~ less than 12800 elements). For large meshes (12800 elements or higher) finite element software FEFLOW performs better. Nevertheless, the study showed that improving the numerical performance by optimizing discretization methods, solvers and parallelization methods still remain a crucial field of research.
Keywords: solute transport modeling, numerical modeling; Tracer test; Geothermal; EGS; solute tracer; sorptive tracer; water fracture; gel-proppant fracture; single-well tests; injection-flowback tests